The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it is designed to process untrusted data from web pages. * Ingestion points: External content is loaded into the agent context via browser navigation tools like page.goto and chrome.navigatePage. * Boundary markers: No explicit instructions are provided to delimit or sanitize web content before processing. * Capability inventory: The agent can interact with elements, fill forms, and execute JavaScript within the page context (page.evaluate). * Sanitization: There is no documentation for sanitizing data retrieved from the web.
[COMMAND_EXECUTION]: The skill includes local automation scripts and instructions for standard CLI usage. * Evidence: The scripts/test-scaffold.ps1 PowerShell script performs local filesystem operations to create a test directory and generate boilerplate files. * Evidence: The documentation instructs the agent on using npx playwright for test execution and code generation.
[EXTERNAL_DOWNLOADS]: The skill utilizes well-known dependencies and tools from official sources. * Evidence: The skill references standard installation of Playwright browser binaries and the @axe-core/playwright package for accessibility testing.

web-testing